Dogs can distinguish between specific negative human facial expressions, brain scans show

New research provides evidence that the canine brain not only separates human facial expressions by positive or negative emotion but also distinguishes between specific negative expressions. These distinct patterns of brain activity suggest that dogs have a nuanced neural representation of human emotions. The findings were recently published in the journal iScience.

Behavioral studies have documented that dogs observe and react to human emotional cues. Canines can distinguish between a smiling face and a neutral one. They often modify their behavior based on the emotional states of the humans around them.

Despite this observed flexibility, the exact brain processes supporting these abilities are not fully understood. Previous neuroimaging work mostly tested positive versus negative emotions or relied on images of the dogs’ caregivers. Because of this, it was not known if canine brains simply register a positive or negative state or if they can process the differences between distinct emotional expressions of the same valence.

To rule out the effect of familiarity, scientists designed a new experiment using only photographs of unfamiliar human faces. They aimed to map the specific neural mechanisms underlying how dogs process happy expressions. They also wanted to observe if the canine brain exhibits unique patterns of activity for different negative facial expressions.

In the first part of the study, the researchers recruited eight awake pet dogs, primarily border collies. The animals were trained to lie completely still inside a functional magnetic resonance imaging scanner. This device measures brain activity by detecting changes in blood flow. During the brain scans, the dogs viewed alternating blocks of photographs showing strangers with either happy or neutral facial expressions.

Free daily newsletter

The researchers matched the images for low-level visual properties like brightness and contrast to ensure the dogs were reacting to the emotions rather than simple lighting differences. The scans indicated that happy faces elicited stronger brain activity than neutral faces. This increased activity occurred in a large cluster located in the right temporal cortex that extended into the caudate nucleus.

The temporal cortex is an area associated with processing visual and social information. The caudate nucleus is a brain region typically linked to the processing of rewards. This pattern suggests that happy human faces, even those of strangers, function as rewarding social stimuli for dogs.

Next, the authors conducted a second experiment to test whether this specific brain region responds uniquely to happiness or just to emotional faces in general. They scanned twelve dogs, including some from the first group. This time, they showed the animals images of unfamiliar humans expressing happiness, anger, fear, and sadness.

The scientists used a machine learning computer model to analyze the brain data, restricting their focus only to the temporal-caudate region identified in the first experiment. The model analyzed the data and successfully distinguished the brain activity patterns of happiness from those of anger, fear, and sadness. The model performed better than chance only when happiness was one of the two emotions being compared.

When the model tried to distinguish between two negative expressions using only this brain region, the results were not statistically significant. This indicates that the right temporal-caudate network has a preferential responsiveness to happy human faces. It does not simply react to any emotional facial expression.

To find out if dogs can differentiate between the negative emotions, the researchers expanded their analysis to the entire brain. They looked for distributed patterns of activity that generalized across different images of the same emotion but varied between different emotions. They also used computational models of the early visual cortex to ensure that subtle visual shapes did not skew the data.

This whole-brain analysis provided evidence that dog brains distinguish between certain negative facial expressions. The researchers found that activity patterns in a region called the right rostral suprasylvian gyrus differentiated sadness from fear. A different set of regions, including the right mid ectosylvian gyrus and left splenial gyrus, differentiated anger from fear.

The scientists did not find distinct brain patterns that could separate sadness from anger. Distinguishing between these two negative emotions may require additional sensory information, such as body language or vocal cues. Overall, the distinct neural patterns related to fearful faces suggest that fear might hold a particular biological relevance for dogs compared to other negative emotions.

Reading these findings requires some caution regarding how canine perception is interpreted. The ability of the canine brain to separate these expressions neurologically does not mean that dogs experience or categorize emotions exactly as humans do. The labels of happiness, fear, anger, and sadness are human descriptions of the stimuli. They do not necessarily reflect a human-like semantic understanding of those feelings in the dogs’ minds.

The study sample also consisted largely of border collies, a breed known for its intense focus on human social cues. This high level of social tuning suggests that the findings might not apply equally to all canine breeds. Free-ranging dogs or breeds less oriented toward human interaction might process facial expressions differently.

The research relied on static photographs rather than real-world emotional events. In everyday life, dogs interpret dynamic facial movements paired with human body language, vocal tone, and scent. Future research plans to use more naturalistic, moving, and multisensory stimuli to observe how dogs integrate all of these social signals.

The study, “Dog brain representations of human facial expressions: Encoding happiness and differentiating specific negative expressions,” was authored by Raúl Hernández-Pérez, Luis Concha, Attila Andics, Rodolfo Bernal-Gamboa, and Laura V. Cuaya.